Can news-based economic sentiment predict bubbles in precious metal markets?

Maghyereh, Aktham; Abdoh, Hussein

doi:10.1186/s40854-022-00341-w

Financial Innovation

Table 9 Probit results using alternative measures of sentiment

From: Can news-based economic sentiment predict bubbles in precious metal markets?

	Dependent variable: Bubble
	Gold	Silver	Palladium	Platinum
Panel A: Consumer Sentiment Index-(MCSI)
\(MCSI_{t - 1}\)	− 0.6237**	− 0.4250	0.0505	− 0.5351***
\(MCSI_{t - 1}\)	(0.0320)	(0.2530)	(0.4390)	(0.0000)
\(Inflation_{t - 1}\)	0.0454**	0.0635*	0.0219***	0.0323***
\(Inflation_{t - 1}\)	(0.0495)	(0.0768)	(0.0000)	(0.0000)
\(US{\text{DI}}_{t - 1}\)	− 0.6837***	− 0.7483**	− 0.6510*	0.1133
\(US{\text{DI}}_{t - 1}\)	(0.0000)	(0.0100)	(0.0970)	(0.2090)
\({\text{EFR}}_{t - 1}\)	− 0.1346*	0.2611	− 0.3583***	− 0.2892***
\({\text{EFR}}_{t - 1}\)	(0.0820)	(0.2310)	(0.0000)	(0.0000)
\(T - Spread_{t - 1}\)	− 0.5169**	− 0.3037**	− 0.2870	− 0.4358**
\(T - Spread_{t - 1}\)	(0.0310)	(0.0100)	(0.5050)	(0.0170)
\({\text{GEA}}_{t - 1}\)	0.0055***	0.0060***	0.0014*	0.0578***
\({\text{GEA}}_{t - 1}\)	(0.0010)	(0.0300)	(0.0680)	(0.0010)
\(Constant\)	0.5464***	0.8085***	0.9020***	0.5881***
\(Constant\)	(0.0021)	(0.0000)	(0.0000)	(0.0000)
\(Observations\)	420	420	420	420
McFadden's pseud-R²	0.6464	0.5775	0.4357	0.5603
Log-likelihood	− 105.8980	− 47.6920	− 88.2729	− 159.7165
Hosmer–Lemeshow test	14.93	10.05	13.65	9.02
Hosmer–Lemeshow test	(0.1019)	(0.2093)	(0.1006)	(0.2290)
Panel B: Sentiment index in Baker and Wurgler (2006)-(SIBW)
\(S{\text{SIBW}}_{t - 1}\)	− 0.7794***	0.1973	− 0.0690	− 0.3565**
\(S{\text{SIBW}}_{t - 1}\)	(0.0060)	(0.7650)	(0.5930)	(0.0320)
\(Inflation_{t - 1}\)	0.2538***	0.0182***	0.0625***	0.0091***
\(Inflation_{t - 1}\)	(0.0068)	(0.0093)	(0.0000)	(0.0000)
\(US{\text{DI}}_{t - 1}\)	− 1.0930***	− 0.9085***	− 1.0120***	0.0358
\(US{\text{DI}}_{t - 1}\)	(0.0000)	(0.0020)	(0.0094)	(0.1310)
\({\text{EFR}}_{t - 1}\)	− 0.1996**	− 0.1831**	− 0.2339***	− 0.8365***
\({\text{EFR}}_{t - 1}\)	(0.0145)	(0.0035)	(0.0000)	(0.0000)
\(T - Spread_{t - 1}\)	− 0.6899***	− 1.1837**	− 0.3009*	− 0.4819*
\(T - Spread_{t - 1}\)	(0.0020)	(0.0180)	(0.0930)	(0.0690)
\({\text{GEA}}_{t - 1}\)	0.0046***	0.0062**	0.0057***	0.0072**
\({\text{GEA}}_{t - 1}\)	(0.0010)	(0.0220)	(0.0000)	(0.0220)
\(Constant\)	0.4209***	0.5323***	0.4170***	0.8651***
\(Constant\)	(0.0060)	(0.0002)	(0.0000)	(0.0000)
\(Observations\)	408	408	408	408
McFadden's pseud-R²	0.6803	0.5721	0.3365	0.4256
Log-likelihood	− 104.1921	− 48.2371	− 161.8798	− 84.6209
Hosmer–Lemeshow test	12.76	4.71	5.19	5.13
Hosmer–Lemeshow test	(0.2371)	(0.8946)	(0.6050)	(0.6106)

This table reports the results using alternative measures of sentiment. Panels A and B report the results using the MCSI and SIBW, respectively. The dependent variable is a binary that equals 1 (bubble dates) and 0 (none-bubble dates) identified by the GSADF procedure. The Hosmer–Lemeshow test is a statistical test for goodness of fit for probit regressions, following the \(\chi^{2}\) distribution. A large \(\chi^{2}\) value (with small p value \(< 0.05\)) indicates poor fit regression model. Robust standard errors are given in parentheses. p values are given in brackets.*, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively.

Back to article page